Towards large scale multi-target tracking

Multi-target tracking is intrinsically an NP-hard problem and the complexity of multi-target tracking solutions usually do not scale gracefully with problem size. Multi-target tracking for on-line applications involving a large number of targets is extremely challenging. This article demonstrates th...

Full description

Bibliographic Details
Main Authors: Vo, Ba-Ngu, Vo, Ba Tuong, Reuter, S., Lam, Q., Dietmayer, K.
Format: Conference Paper
Published: SPIE 2014
Online Access:http://hdl.handle.net/20.500.11937/20355
Description
Summary:Multi-target tracking is intrinsically an NP-hard problem and the complexity of multi-target tracking solutions usually do not scale gracefully with problem size. Multi-target tracking for on-line applications involving a large number of targets is extremely challenging. This article demonstrates the capability of the random finite set approach to provide large scale multi-target tracking algorithms. In particular it is shown that an approximate filter known as the labeled multi-Bernoulli filter can simultaneously track one thousand five hundred targets in clutter on a standard laptop computer. © 2014 SPIE.